OpenAI
Configure an OpenAI API key and model in K8Studio settings when you want hosted reasoning for troubleshooting, YAML review, and operational guidance.
K8Studio learning resources
AI assistance for Kubernetes operations
K8Studio Copilot helps DevOps engineers understand cluster state, debug workloads, review manifests, and prepare safer Kubernetes changes from inside the desktop IDE. You choose the model provider: OpenAI, Anthropic Claude, or a local Ollama model.
How Copilot is different
Step 1
It starts from your current context
Copilot can use what you have selected in K8Studio: cluster, namespace, workload, YAML, logs, events, metrics, and visible resource details.
Step 2
It reasons with Kubernetes structure
Instead of treating Kubernetes as plain text, Copilot can look at resources, relationships, events, rollout state, and manifests through K8Studio.
Step 3
It uses tools for real actions
When the request is operational, Copilot can use K8Studio MCP tools to navigate, inspect, analyze, and prepare a concrete Kubernetes action.
Step 4
You review mutations before apply
For create, patch, update, delete, scale, and restart actions, K8Studio requires confirmation and runs safety checks before applying changes.
Model choice
Switch between hosted and local AI models
K8Studio keeps the provider choice in application settings. Teams can use hosted APIs for stronger managed models or point Copilot at a local Ollama server when they need stricter data boundaries.
Claude
Configure an Anthropic API key and Claude model when you want long-context explanations, careful review, and detailed incident analysis.
Ollama
Configure an Ollama URL and local model when you want Copilot to talk to a model running on your machine or inside a controlled network.
Where you change models
Open K8Studio settings, choose the AI Agent provider, then enter either an API key for OpenAI or Anthropic, or an Ollama URL and model name for local inference. The same Copilot UI can work across providers, so teams can standardize on the model policy that fits each environment.
K8Studio MCP
MCP is the tool layer behind Copilot actions
MCP, the Model Context Protocol, gives the assistant structured tools instead of asking it to guess from prose. K8Studio exposes MCP tools for the current application and Kubernetes context so Copilot can take precise, reviewable steps.
Navigate K8Studio
Open cluster pages, resource grids, details, logs, YAML, events, metrics, Helm, RBAC, terminal, and security views.
Inspect Kubernetes resources
List, read, and summarize resources using structured arguments instead of invented kubectl commands or free-form guesses.
Analyze failures
Review selected resources, logs, events, rollout state, and related objects to explain what is wrong and what to check next.
Prepare changes
Create, patch, update, delete, scale, or restart resources through guarded MCP tools when the user clearly asks for a change.
Guardrails built into the workflow
AI assistance should make Kubernetes work faster without turning production changes into blind automation. K8Studio keeps human review and cluster authorization in the path.
Mutating MCP tools require K8Studio confirmation.
K8Studio runs RBAC and server-side dry-run preflight before applying Kubernetes changes.
Copilot should use read-only tools first for troubleshooting and only mutate when the user asks clearly.
Ollama lets teams keep model traffic local when they do not want hosted AI endpoints.
Ask better Kubernetes questions
Try questions like: why is this deployment not ready, summarize these logs, explain this YAML, compare requested and actual resources, show me related services, or prepare a patch for this container image.